IoT Botnet Threat Detection and Classification: A Binary Class Approach
Maisha Maliha, Vaishnavi Satya Sreeja Ankam, Nagamani Rudraraju, Wassnaa Al-Mawee · 2024
In the ever-expanding Internet of Things (IoT) landscape, security holds topmost importance. In this study, our focus revolves around the significance of securing IoT devices against potential threats, particularly the crucial task of botnet detection. Employing the N aiBot dataset-a labeled dataset for binary classification-we approached this supervised learning problem using Machine Learning(ML) algorithms, including Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forest (RF), Decision Tree (DT), and XGBoost. Integrating Principal Component Analysis (PCA) and feature selection techniques, we assessed our models using a variety of evaluation metrics, including Receiver Oper-ating Characteristic (ROC) analysis. Additionally, we computed false alarm and missed detection rates to evaluate the models' performance. Our study contributes to enhancing IoT security by providing a comprehensive evaluation of Machine Learning techniques for botnet detection.